AIMS Mathematics, 4(3): 779–791. DOI:10.3934/math.2019.3.779 Received: 09 April 2019 Accepted: 02 June 2019 http://www.aimspress.com/journal/Math Published: 05 July 2019 Research article The new class Lz;p;E of s− type operators Pınar Zengin Alp and Emrah Evren Kara∗ Department of Mathematics, Duzce¨ University, Konuralp, Duzce, Turkey * Correspondence: Email: [email protected]. Abstract: The purpose of this study is to introduce the class of s-type Z u; v; lp (E) operators, which we denote by Lz;p;E (X; Y), we prove that this class is an operator ideal and quasi-Banach operator ideal by a quasi-norm defined on this class. Then we define classes using other examples of s-number sequences. We conclude by investigating which of these classes are injective, surjective or symmetric. Keywords: block sequence space; operator ideal; s-numbers; quasi-norm Mathematics Subject Classification: 47B06, 47B37, 47L20. 1. Introduction In this study, the set of all natural numbers is represented by N and the set of all nonnegative real numbers is represented by R+ . If the dimension of the range space of a bounded linear operator is finite, it is called a finite rank operator [1]. Throughout this study, X and Y denote real or complex Banach spaces. The space of all bounded linear operators from X to Y is denoted by B (X; Y) and the space of all bounded linear operators from an arbitrary Banach space to another arbitrary Banach space is denoted by B. The theory of operator ideals is a very important field in functional analysis. The theory of normed operator ideals first appeared in 1950’s in [2]. In functional analysis, many operator ideals are constructed via different scalar sequence spaces. An s- number sequence is one of the most important examples of this. The definition of s- numbers goes back to E. Schmidt [3], who used this concept in the theory of non-selfadjoint integral equations. In Banach spaces there are many different possibilities of defining some equivalents of s- numbers, namely Kolmogorov numbers, Gelfand numbers, approximation numbers, and several others. In the following years, Pietsch give the notion of s- number sequence to combine all s- numbers in one definition [4–6]. A map S : K ! (sr(K)) 780 which assigns a non-negative scalar sequence to each operator is called an s-number sequence if for all Banach spaces X; Y; X0 and Y0 the following conditions are satisfied: (i) kKk = s1 (K) ≥ s2 (K) ≥ ::: ≥ 0, for every K 2 B (X; Y) ; (ii) sp+r−1 (L + K) ≤ sp (L) + sr (K) for every L; K 2 B (X; Y) and p; r 2 N; (iii) sr (MLK) ≤ kMk sr (L) kKk for all M 2 B (Y; Y0) ; L 2 B (X; Y) and K 2 B (X0; X) ; where X0; Y0 are arbitrary Banach spaces, (iv) If rank (K) ≤ r; then sr (K) = 0; n (v) sn−1 (In) = 1; where In is the identity map of n-dimensional Hilbert space l2 to itself [7]. sr (K) denotes the r − th s−number of the operator K. Approximation numbers are frequently used examples of s-number sequence which is defined by Pietsch. ar (K), the r-th approximation number of a bounded linear operator is defined as ar (K) = inf f kK − Ak : A 2 B (X; Y) , rank (A) < r g ; where K 2 B (X; Y) and r 2 N [4]. Let K 2 B (X; Y) and r 2 N: The other examples of s-number 0 sequences are given in the following, namely Gel f and number (cr (K)), Kolmogorov number (dr (K)), Weyl number (xr (K)), Chang number (yr (K)) ; Hilbert number (hr (K)) ; etc. For the definitions of these sequences we refer to [1]. In the sequel there are some properties of s−number sequences. When any isometric embedding J 2 B (Y; Y0) is given and an s-number sequence s = (sr) satisfies sr (K) = sr (JK) for all K 2 B (X; Y) the s-number sequence is called injective [8, p.90]. Proposition 1. [8, p.90–94] The number sequences (cr (K)) and (xr (K)) are injective. When any quotient map S 2 B (X0; X) is given and an s-number sequence s = (sr) satisfies sr (K) = sr (KS) for all K 2 B (X; Y) the s-number sequence is called surjective [8, p.95]. Proposition 2. [8, p.95] The number sequences (dr (K)) and (yr (K)) are surjective. Proposition 3. [8, p.115] Let K 2 B (X; Y). Then the following inequalities hold: i)h r (K) ≤ xr (K) ≤ cr (K) ≤ ar (K) and ii)h r (K) ≤ yr (K) ≤ dr (K) ≤ ar (K) : Lemma 1. [5] Let S; K 2 B (X; Y) ; then jsr (K) − sr (S )j ≤ kK − S k for r = 1; 2;:::: Let ! be the space of all real valued sequences. Any vector subspace of ! is called a sequence space. In [9] the space Z u; v; lp is defined by Malkowsky and Savas¸as follows: 8 1 n p 9 > X X > Z u; v; l = <x 2 ! : u v x < 1= p > n k k > : n=1 k=1 ; where 1 < p < 1 and u = (un) and v = (vn) are positive real numbers. The Cesaro sequence space cesp is defined as ( [10, 11, 19]) 8 1 0 n 1p 9 > X B1 X C > ces = <x = (x ) 2 ! : B jx jC < 1= ; 1 < p < 1: p > k @Bn k AC > : n=1 k=1 ; AIMS Mathematics Volume 4, Issue 3, 779–791. 781 1 P p If an operator K 2 B (X; Y) satisfies (an (K)) < 1 for 0 < p < 1, K is defined as an lp type n=1 operator in [4] by Pietsch. Afterwards ces-p type operators which is a new class obtained via Cesaro sequence space is introduced by Constantin [12]. Later on Tita in [14] proved that the class of lp type operators and ces-p type operators coincide. (s) In [15], &p ; the class of s−type Z u; v; lp operators is given. For more information about sequence spaces and operator ideals we refer to [1, 13, 16, 18, 20]. Let X0 , the dual of X; be the set of continuous linear functionals on X: The map x∗ ⊗ y : X ! Y is defined by (x∗ ⊗ y)(x) = x∗ (x) y where x 2 X, x∗ 2 X0 and y 2 Y. A subcollection = of B is said to be an operator ideal if for each component = (X; Y) = = \ B (X; Y) the following conditions hold: (i) if x∗ 2 X0 , y 2 Y, then x∗ ⊗ y 2 = (X; Y) ; (ii) if L; K 2 = (X; Y) ; then L + K 2 = (X; Y) ; (iii) if L 2 = (X; Y) ; K 2 B (X0; X) and M 2 B (Y; Y0) ; then MLK 2 = (X0; Y0) [6]. Let = be an operator ideal and ρ : = ! R+ be a function on =. Then, if the following conditions hold: (i) if x∗ 2 X0 , y 2 Y, then ρ (x∗ ⊗ y) = kx∗k kyk ; (ii) if 9C ≥ 1 such that ρ (L + K) ≤ C ρ (L) + ρ (K) ; (iii) if L 2 = (X; Y) ; K 2 B (X0; X) and M 2 B (Y; Y0) ; then ρ (MLK) ≤ kMk ρ (L) kKk, ρ is said to be a quasi-norm on the operator ideal = [6]. For special case C = 1, ρ is a norm on the operator ideal =: If ρ is a quasi-norm on an operator ideal =, it is denoted by =; ρ. Also if every component = (X; Y) is complete with respect to the quasi-norm ρ, =; ρ is called a quasi-Banach operator ideal. Let =; ρ be a quasi-normed operator ideal and J 2 B (Y; Y0) be a isometric embedding. If for every operator K 2 B (X; Y) and JK 2 = (X; Y0) we have K 2 = (X; Y) and ρ (JK) = ρ (K), =; ρ is called an injective quasi-normed operator ideal. Furthermore, let =; ρ be a quasi-normed operator ideal and S 2 B (X0; X) be a quotient map. If for every operator K 2 B (X; Y) and KS 2 = (X0; Y) we have K 2 = (X; Y) and ρ (KS) = ρ (K), =; ρ is called an surjective quasi-normed operator ideal [6]. Let K0 be the dual of K. An s− number sequence is called symmetric (respectively, completely 0 0 symmetric) if for all K 2 B, sr (K) ≥ sr(K ) (respectively,sr (K) = sr(K )) [6]. 0 Lemma 2. [6] The approximation numbers are symmetric, i.e., ar(K ) ≤ ar(K) for K 2 B . Lemma 3. [6] Let K 2 B. Then 0 0 cr(K) = dr(K ) and cr(K ) ≤ dr(K): 0 In addition , if K is a compact operator then cr(K ) = dr(K) . AIMS Mathematics Volume 4, Issue 3, 779–791. 782 Lemma 4. [8] Let K 2 B. Then 0 0 xr(K) = yr(K ) and yr(K ) = xr(K) . The dual of an operator ideal = is denoted by =0 and it is defined as [6] 0 n 0 0 o = (X; Y) = K 2 B (X; Y) : K0 2 = Y ; X . An operator ideal = is called symmetric if = ⊂ =0 and is called completely symmetric if = = =0 [6]. Let E = (En) be a partition of finite subsets of the positive integers which satisfies max En < min En+1 + for n 2 N : In [21] Foroutannia defined the sequence space lp (E) by 8 p 9 > X1 X > <> => lp (E) = x = (xn) 2 ! : x j < 1 ; (1 ≤ p < 1) :> ;> n=1 j2En with the seminorm kj·jkp;E ; which defined as: p 1 0 1 1 p BX X C kjxjk = B x C : p;E @B j AC n=1 j2En For example, if En = f3n − 2; 3n − 1; 3ng for all n, then x = (xn) 2 lp (E) if and only if 1 P p jx3n−2 + x3n−1 + x3nj < 1: It is obvious that kj·jkp;E is not a norm, since we have kjxjkp;E = 0 while n=1 + x = (−1; 1; 0; 0;:::) and En = f3n − 2; 3n − 1; 3ng for all n.